Edouard Grave
Equal-contribution author of Meta's LLaMA paper; co-author of fastText and of Kyutai's Moshi
The November 2023 announcement of the Kyutai research lab describes Edouard Grave as a research scientist with experience in the corporate research labs Apple MLR and Facebook AI Research (FAIR), and says that at FAIR, Grave co-led the development of the LLaMA large language model and the fastText library. Grave is one of the equal-contribution authors of LLaMA: Open and Efficient Foundation Language Models (2023) and a co-author of the two 2016 fastText papers on subword word vectors and efficient text classification. Kyutai's team page lists Grave among its alumni with the title Chief Language Officer, and Grave co-authored Kyutai's Moshi speech-text foundation model paper (2024).134527
Work
LLaMAPublished research
One of five equal-contribution authors of LLaMA: Open and Efficient Foundation Language Models (2023), whose author list gives a Meta AI affiliation; the Kyutai launch announcement says Grave co-led the development of LLaMA at FAIR.31
fastText (external site: github.com)Open-source code
Co-author of Enriching Word Vectors with Subword Information (2016) and Bag of Tricks for Efficient Text Classification (2016); the fastText repository is published under the MIT License, and the Kyutai launch announcement says Grave co-led the library's development.4561
Moshi (external site: arxiv.org)Published research
Co-author of Moshi: a speech-text foundation model for real-time dialogue (2024), which introduces a full-duplex spoken dialogue framework and points to the kyutai-labs/moshi repository.7
Other documented roles
Roles the sources document but that are not confirmed as current.
- Chief Language Officer (listed among alumni), Kyutai21
- Research scientist; co-led the development of LLaMA and the fastText library, Meta (Facebook AI Research, FAIR)13
- Research scientist in the Apple MLR corporate research lab, Apple (Apple MLR)1
This profile covers public professional work only, from the sources below. It is not an endorsement or a ranking. To correct or remove it, use the correction form.